Molecular beacons with oxidized bases report on substrate specificity of DNA oxoguanine glycosylases

DNA glycosylase enzymes recognize and remove structurally distinct modified forms of DNA bases, thereby repairing genomic DNA from chemically induced damage or erasing epigenetic marks. However, these enzymes are often promiscuous, and advanced tools are needed to evaluate and engineer their substrate specificity. Thus, in the present study, we developed a new strategy to rapidly profile the substrate specificity of 8-oxoguanine glycosylases, which cleave biologically relevant oxidized forms of guanine. We monitored the enzymatic excision of fluorophore-labeled oligonucleotides containing synthetic modifications 8-oxoG and FapyG, or G. Using this molecular beacon approach, we identified several hOGG1 mutants with higher specificity for FapyG than 8-oxoG. This approach and the newly synthesized probes will be useful for the characterization of glycosylase substrate specificity and damage excision mechanisms, as well as for evaluating engineered enzymes with altered reactivities.


Introduction
DNA is constantly altered by epigenetic modications and chemical exposures. Sophisticated molecular mechanisms involving epigenetic writers/erasers and repair enzymes have evolved, therefore, to manipulate such structures. 1,2 Relying on many of these enzymes, modern DNA manipulation and gene editing technologies are transforming our understanding of the genome and how to treat diseases, such as involving zinc-nger nucleases, 3 transcription activator-like effector nucleases (TAL-ENS), 4 clustered regularly interspaced short palindromic repeat (CRISPR) 5 systems, base editors, 6 and glycosylases. 7 However, many of these enzymes are inherently promiscuous, limiting their specicity to experimentally manipulate DNA on the basis of particular structural modications in DNA. 8,9 For example, DNA glycosylases are efficient base-excision enzymes that remove nucleobases with diverse small modications, such as those resulting from oxidation and methylation of DNA. Thus, new chemical and enzymological strategies for dening and altering the substrate specicity of glycosylases are expected to advance precision DNA manipulation aer recognizing chemical modications.
DNA base oxidation, arising from reactions with reactive oxygen species (ROS) resulting from environmental chemical exposures, UV or ionizing irradiation, and cellular metabolism, 10 is a major form of DNA chemical modication, signicantly impacting genome integrity and cell function. 11 Of the canonical bases in DNA, guanine is the most easily oxidized, due to its low redox potential amongst the four DNA bases. 12 Several structurally distinct oxidation products result with potentially distinct genomic distribution and biological impacts. 13 Major products include 8-oxoguanine (8-oxoG) and 2,6-diamino-4-oxo-5-formamidopyrimidine (FapyG) (Fig. 1). 14 Error-prone replication of 8-oxoG or FapyG induces G to T transversion and G to A transition mutations. 15 glycosylases that remove these oxidized nucleobases initiate base excision repair, 13 and are key reagents for emerging DNA damage detection and sequencing technologies. 7,[17][18][19][20][21][22] Oxoguanine glycosylases have been used widely to detect DNA oxidation, including in the comet assay, 19 mass spectrometry 20 and DNA damage sequencing. 18,21,22 In the oxidationsequencing methods click-code-seq and entrap-seq, formamidopyrimidine-DNA glycosylase (Fpg) or a human 8-oxoG glycosylase (hOGG1) K249Q mutant were used to enrich and map guanine oxidation in yeast and mouse genomes, respectively. 18,22 While the majority of mapped sites were expected to be 8-oxoG due to its high prevalence in the genome, the general approach is actually limited to mapping the oxidized base substrate scope of the glycosylase used, and the distribution of specic chemical forms of oxidation products could not be resolved, such as distinguishing between 8-oxoG vs. FapyG. As a result, there are no high specicity DNA oxidation maps available, and there is a need to rapidly assess the substrate specicity of enzyme variants in order to engineer enzymes with substrate scopes different from what is found in nature.
The most common way to characterize glycosylase activity is separating and imaging cleaved DNA by gel electrophoresis, 23 a method with insufficient throughput for screening enzyme variants. Likewise, LC-MS and qPCR can be used but involve tedious sample preparation. 24,25 Fluorescent molecular beacons have been established as excellent tools for characterizing glycosylase activity in a simple and real-time manner. [26][27][28][29] However, to our knowledge, such a strategy has not been established as yet to evaluate glycosylase substrate specicity and screen mutants with altered function due to a lack of synthetic probes containing different glycosylase substrates and the typical use of puried glycosylase enzymes.
In this study, we developed a three-color molecular beacon system to evaluate the 8-oxoG vs. FapyG specicity of glycosylase enzymes. The goal was to compare enzyme activity using oligonucleotide substrates containing 8-oxoG or FapyG, or a control oligonucleotide containing G (Fig. 2). Three molecular beacons were created for this platform, each with a different uorophore that is released as an indicator of the relative efficiency of glycosylase activity on the corresponding base structure. This novel molecular beacon platform was used to characterize the specicity of hOGG1 and several hOGG1 variants in crude whole-cell lysates. Furthermore, we evaluated the substrate specicity of a random hOGG1 D322 variant library and identied variants with an increased relative capacity to excise FapyG. This approach has the capacity to report on glycosylase substrate specicity directly in cell lysates by using orthogonal uorescent reporters, offering the possibility for high-throughput glycosylase variant screening.

Synthesis of three-color molecular beacons
As a rst step to create the three-color molecular beacon system to evaluate the specicity of glycosylase enzymes for 8-oxoG vs. FapyG excision, we synthesized molecular beacons containing competing substrates at a dened position, and also equipped them with a uorophore and quencher pair. We employed a post-synthetic double modication strategy whereby a nitropyrimidine (NP-dG) used previously as a FapyG precursor 30 (Schemes 1I; S1 and S6 †) and a 5 0 amino guanosine (AM-dG) phosphoramidite (Schemes S2 and S10 †), were synthesized and incorporated into the 10mer oligonucleotide 5 0 -AM-dGGTCTNP-dGATGG-3 0 (Scheme 1II). We found that using AM-dG 31 for the attachment of FAM provided an $80-fold higher signal-to-noise ratio than using a commercial C6 amino modier (Fig. S1 †). Initial attempts to catalyze the reduction of the nitro group (Scheme 1, step 2) with palladium on carbon 30 lead to the Scheme 1 Synthesis of a FapyG-containing molecular beacon.
cleavage of oligonucleotides at the NP-dG site and very low product yields; this problem was circumvented by using NaBH 4 and a nickel boride catalyst (Fig. S2 †). 32 Subsequent steps included deprotection of the MMT group with 80% acetic acid and installation of the Cy3 uorophore on the 5 0 end of the oligonucleotide by NHS ester chemistry (Scheme 1, step 4). Finally, the full-length molecular beacon was obtained by ligating the FapyG-containing 10mer oligonucleotide with a BHQ2-modied oligonucleotide (Scheme 1, step 5; Fig. S3 †) and purication on a polyacrylamide gel. As in previous studies involving a Fapy-containing oligonucleotide, the modication exists as a mixture of a and b anomers. [33][34][35][36] Finally, 8-oxoG-and G-containing molecular beacons were synthesized in the same manner as FapyG (experimental section).

Optimization and validation of molecular beacon-based assay
Having color-coded 8-oxoG, FapyG and G molecular beacon substrates in hand, we tested the relative cleavage rates in the presence of enzymes: Fpg, hOGG1, and apurinic/apyrimidinic endonuclease (APE1). Fpg efficiently cleaved the molecular beacons containing FapyG (Cy3) and 8-oxoG (Cy5), yielding Cy3and Cy5-labeled 3 0 -phosphated short oligonucleotides (Fig. 3, lane 2). Reactions catalyzed by hOGG1 and APE1 yielded 3 0hydroxyl oligonucleotides (Fig. 3, lane 4). The identity of the 5mer products arising from the cleavage catalyzed by Fpg . By evaluating the reaction products by mass spectrometry, we determined that the increase in the FAM signal was caused by the cleavage of the BHQ1 quencher at the 3 0 end of the molecular beacon ( Fig. S7 †), possibly due to the 3 0 -diesterase activity of APE1. This effect was observed for BHQ1 but not for BHQ2. 38,39 To avoid quencher cleavage, N,N-dimethylethylenediamine (DMEDA), which was demonstrated previously to promote abasic site cleavage, 40 was used in place of APE1. Replacing APE1 with DMEDA in the hOGG1 excision reaction also was effective for cleaving 8-oxoG-and FapyG-containing molecular beacons, producing mainly b-elimination products (Fig. 3, lane 5). 40 Furthermore, a linear correlation was observed between the uorescence signal resulting from Cy5 and the concentration of the molecular beacon containing 8-oxoG (concentration range 50-500 nM, R 2 ¼ 0.994, Fig. S8 †). The outcome of these studies was an optimized robust strategy to compare the specicity of puried glycosylases for the cleavage of 8-oxoG vs. FapyG.

Glycosylase activity evaluation directly in crude cell lysates
For rapid proling of several glycosylase variants, such as those generated in the course of protein engineering, the strategy should also effectively report on enzyme specicity from cell lysates rather than puried enzymes. Therefore, EDTA was added to the mixture to sequester magnesium and inhibit exonuclease (Exo I & III) activity without reducing hOGG1 or Fpg function (Fig. 4). Meanwhile, less than 3% cleavage was observed in assays with hOGG1, DMEDA, APE1, uracil-DNA glycosylase (UDG) or human alkyl adenine DNA glycosylase (hAAG). Furthermore, we transformed a hOGG1 plasmid into E. coli cells and overexpressed hOGG1. Direct analysis of these cell lysates led to $80% cleavage of the molecular beacons within 30 min, compared to 2-5% cleavage by lysate from cells transformed with an empty vector (Fig. S9 †). Thus, we could evaluate the glycosylase specicity in E. coli cell lysates without protein purication.

Substrate specicity proling of known hOGG1 variants and single mutation variant libraries
With a robust in situ glycosylase proling assay established, we set out to characterize the inuence of particular amino acid residues in the substrate specicity of hOGG1. Thus, hOGG1 single-nucleotide polymorphic variants (A288V, D322N and S326C) and phosphorylation mimics (S231E, S232E, S280E and S326E), previously characterized by using traditional gel-based assays, 23,41,42 were proled with regards to relative initial rates of enzymatic excision ( Fig. 5 and S10 †) and substrate specicities (Table 1) using the molecular beacon platform. Variants with mutations more distal to the catalytic center (S326C, S326E, S231E, S232E and S280E) had similar FapyG/8-oxoG specicity ratios to the wild type enzyme, ranging between 0.85 and 1.33fold (Table 1 and Fig. 6). However, the D322N mutant, altered close to the catalytic center, had the highest substrate specicity (3.3-fold) for FapyG. These results demonstrated the simplicity and robustness of the molecular beacon strategy and suggested a candidate residue for further tuning glycosylase specicity.
In the crystal structure of hOGG1, 43 either in the apo form or bound to 8-oxoG, the carboxyl group of D322 forms a hydrogen bond with the imidazole ring of H270. In particular, H270 forms a hydrogen bond with C8]O of the extruded 8-oxoG, serving as a crucial residue during 8-oxoG recognition and repair. 43,44 Thus, we proled the specicity of a hOGG1 D322 variant library derived from 48 colonies generated via site-directed mutagenesis, harbouring 14 different single mutations at site 322 (Fig. 7). Variants D322H, D322S, D322N and D322Q, which have a hydrogen bond acceptor at site 322, had a similar specicity to wild type hOGG1. On the other hand, variants D322R, D322C, D322G, D322L, D322V, D322I, D322K and D322F, which lack hydrogen-bond-acceptor capacity were more than 4-fold selective for FapyG over 8-oxoG. Similar specicity was observed for  a Values are initial rates relative to wild type hOGG1, and were calculated on the basis of change in uorescence emission intensity per min during the rst 8 min of reaction. Data are in Fig. 5 and S10.  D322T, which has an amino and hydroxyl group, and D322Y that has an aromatic side chain. The truncated protein (residues 1-321) and D322P mutant did not excise either FapyG or 8-oxoG (Fig. S11 †). Proline (P) is known to perturb the a helix structure, and therefore it may change the global structure of hOGG1, leading to the observed loss of activity. 45 In general, hOGG1 variants containing hydrophobic amino acids at site 322 appear to have better excision specicity for FapyG over 8-oxoG than hydrophilic amino acids, except for positively charged arginine (R) and lysine (K) (Fig. 7). A caution in interpreting the specicity data is potential conformational effects on FapyG cleavage, since it exists as equilibrating anomers. [33][34][35][36] Finally, these results suggest that the properties of the amino acid residue at position 322, including hydrogen donor/acceptor, hydrophobicity, stereochemistry, side chain size and charge state, inuence the substrate specicity of hOGG1. As the crystal structure of hOGG1 bound to FapyG has not been elucidated, further studies are needed to clarify the role of D322 in the excision of FapyG as compared to 8-oxoG. C253 is another important residue for 8-oxoG cleavage by hOOG1. It forms a dipole-dipole interaction with the K249 residue to sandwich the extruded 8-oxoG. 46,47 Therefore, we also generated several C253 hOGG1 variants and evaluated their substrate selectivity (Fig. S11 †) in the same fashion as described for D322 variants (Fig. 7). Mutation of C253 led to loss of hOGG1 activity for excision of 8-oxoG, suggesting that the presence of aliphatic residues at site 253 of hOGG1 blocks the entrance of 8-oxoG, consistent with previous observations of low activity of variant C253L. 48 On the other hand, mutants C253M and C253L retained activity for FapyG excision, with more than 10-fold specicity for FapyG over 8-oxoG. These results suggest that a large residue at site 253 signicantly impedes 8-oxoG excision, but has little effect on FapyG excision, potentially due to its greater structural exibility. These observations suggest that despite the similarities of 8-oxoG and FapyG, there are features of their interactions with glycosylases that can allow them to be distinguished. Nonetheless, extensive further screening of other single-or multi-mutation variants will be required for very large gains in substrate selectivity. Finally, the results from screening D322 and C253 mutants demonstrate the potency of the molecular beacon method for this purpose.

Conclusions
In conclusion, we developed a three-color molecular beacon platform that can be used to evaluate the substrate specicity of 8-oxoG glycosylases for excision of 8-oxoG vs. FapyG. The approach was optimized to prole E. coli cell lysates overexpressing hOGG1, thus providing a simple and rapid assay for screening glycosylase activity and specicity that does not require protein purication. Using this approach, we identied several D322 and C253 variants with higher FapyG/8-oxoG specicity than the wild type enzyme. These residues appear to help stabilize the interaction between hOGG1 and the oxidized base. It is anticipated that the strategy and new synthetic molecular beacons reported here will enable further development and understanding of 8-oxoG glycosylase function and substrate scope. Furthermore, the simple and modular strategy for rapidly proling glycosylase substrate specicity may be used with any modications incorporated into the beacons to understand and engineer the specicity of diverse DNA-cleaving enzymes.

Materials and reagents
Solvents and chemical reagents were purchased from Sigma Aldrich if not specically mentioned otherwise. Commercial enzymes T4 ligase, exonuclease I, exonuclease III, Fpg, APE1, hAAG and UDG were purchased from New England Biolabs and hOGG1 was purchased from R&D system.

Synthesis of oligonucleotides
Oligonucleotides were synthesized on a MerMade 4 DNA/RNA synthesizer (BioAutomation Corporation, USA) with reagents from Glen Research (USA). Oligonucleotides containing 8-oxoguanine and guanine were synthesized in DMT-off mode and oligonucleotides containing the nitro-precursor (Scheme 1(II)) were synthesized in DMT-on mode. The subsequent resin cleavage and deprotection was carried out in concentrated ammonium hydroxide at room temperature for 24 h. The deprotection of oligonucleotides containing 8-oxoguanine was conducted with 2mercaptoethanol (0.25 M) in ammonium hydroxide.

Post-synthesis of oligonucleotides containing FapyG
Approximately 100 OD 260 of a mixture of oligodeoxynucleotides containing a nitro-precursor (Scheme 1(II)) was dissolved in 490 mL of water, followed by the addition of 10 mL of triethylamine. 1 mg freshly prepared Ni 2 B nanoparticles 32 was added to the mixture and stirred at room temperature. A total of 10 mg sodium borohydride was added to the mixture in portions over the course of 10 min, and then the nanoparticles were removed by ltration of the mixture through a 0.2 mm centrifugal lter (VWR). The desired hydrogenated product (Scheme 1(III)) was puried on a Sep-Pak C18 Classic cartridge with 360 mg sorbent (Waters) and eluted using acetonitrile : water (50 : 50). The resulting oligodeoxynucleotide solution was reduced to a volume of 500 mL using a Speed-Vac concentrator (Genevac Ltd, UK). Then 25 mL of 2-mercaptoethanol and 25 mL of triethylamine were added to the mixture, followed by 1 mL of freshly prepared formylimidazole 30,49 (1 M in THF), which was added in portions over the course of 2 h. When the aminopyrimidine oligodeoxynucleotides had completely disappeared, as conrmed by MS analysis, the reaction mixture was evaporated to remove organic solvents and puried on a Sep-Pak C18 Classic cartridge with 360 mg sorbent.
Subsequently, the puried and lyophilized FapyG-containing oligonucleotides (Scheme 1(IV)) were dissolved in 20 mL 80% acetic acid and shaken at room temperature for 20 min. Aer removing acetic acid with a Speed-Vac concentrator, the MMT-off oligonucleotides were re-dissolved in ice-cold 200 mL sodium bicarbonate solution (0.1 M in H 2 O). The amount of oligonucleotides was measured using a Nanodrop spectrophotometer 2000 (Thermo Fisher Scientic, USA) and about 8-fold excess of Cy3 NHS ester (10 mg mL À1 in DMSO, lumiprobe, USA) was added. The reaction was shaken at room temperature and monitored by MS until the starting material disappeared. The excessive dye was removed using a centrifuge lter (MWCO. 3000, Amicon) and the remaining mixture was puried by high performance liquid chromatography (1260 Innity, Agilent, USA) using a C18 reverse phase column (4.6 Â 250 mm, Phenomenex, USA) with a linear gradient of acetonitrile, 0-40%, for 30 min in 0.05 M triethylammonium acetate (TEAA, pH 7.0). The desired oligonucleotides were lyophilized using a freeze dryer (Labconco Corporation, USA), and quantied using a Nanodrop spectrophotometer and characterized by MS (calc.: 3555, found: 3554).

Post-uorescent labelling of oligonucleotides containing 8oxoguanine and guanine
The deprotected oligonucleotides were re-dissolved in 200 mL sodium bicarbonate solution (0.1 M in H 2 O). The mixture was kept on ice, and about 8-fold excess of dye NHS ester (10 mg mL À1 in DMSO, lumiprobe, USA) was added, Cy5 for 8-oxoG and FAM for G. The reaction was shaken at room temperature and monitored by MS until the starting material disappeared. The excessive dye was removed using a centrifuge lter (MWCO. 3000, MilliporeSigma, USA) and the remaining mixture was puried using a C18 reverse phase column (4.6 Â 250 mm, Phenomenex, USA) on a high performance liquid chromatograph (1260 Innity, Agilent, USA) with a linear gradient of acetonitrile, 0-40%, for 30 min in 0.05 M triethylammonium acetate (TEAA, pH 7.0). The desired oligonucleotides were lyophilized and quantied using a Nanodrop spectrophotometer characterized by MS. Oligonucleotides containing 8-oxoG, calc.: 3580, found: 3577; oligonucleotides containing G: calc.: 3454, found: 3454. Deconvolution of multiple charged ESI peaks was performed with MagTran 1.03. 50

Assembly of molecular beacons
The puried uorophore-labelled oligonucleotides were annealed with 1.2-equivalents of quencher-labelled oligonucleotides (Eurogentec, Belgium) in 1Â cutsmart buffer (New England Biolabs, USA). The resulting mixture was heated at 70 C for 5 min and slowly cooled to 4 C (1 C per min). ATP (nal concentration of 1 mM) and T4 ligase (2000 U) were added and allowed to react at room temperature for 2 h. The crude oligonucleotides were loaded onto 20% denaturing polyacrylamide gels (7 M urea). The gels were cooled to 4 C during electrophoresis (250 V, 1 h). The desired gel bands were cleaved and extracted in 1Â TBE buffer at 4 C overnight in a dark room. The molecular beacons were ltered using a 0.2 mm centrifuge lter, desalted using an MWCO. 3000 centrifuge lter, quantied using a nanodrop, conrmed by MS (Table S1 †) and stored at À20 C until further use.

Enzymatic assays with molecular beacons
The uorescence-based assays were conducted in low-volume 384-well plates (Corning, USA) with 10 mL as the nal reaction volume and 3 mL mineral oil layered on top to avoid evaporation. All the molecular beacons were annealed in 1Â reaction buffer (50 mM potassium acetate, 20 mM tris-acetate, 1 mM EDTA, 100 mg mL À1 BSA, pH 7.9) before use, heating the mixture at 70 C for 5 min and slowly cooled to 4 C (1 C per min). DMEDA was prepared as a 1 M solution (10Â stock) in water and adjusted to pH 7.9 by using acetic acid. The molecular beacon probes (0.5 mM) in buffer (50 mM potassium acetate, 20 mM tris-acetate, 100 mg mL À1 BSA, 1 mM EDTA, 100 mM DMEDA, pH 7.9) were added to wells and mineral oil was gently added to the top of the probe/ buffer mixture and the plates were kept at 37 C for 5 min. Enzymes exonuclease I (0.01 U), exonuclease III (0.1 U), hOGG1 (0.5 pmol), Fpg (0.25 pmol), APE1 (0.25 pmol), hAAG (0.05 U) or UDG (0.05 U) were added to each well, and uorescence was monitored with a Tecan innite 2000 plate reader at 37 C for 60 min. Monitored excitation and emission wavelengths included the FAM channel: Ex 490 nm/9, Em 520 nm/20; Cy3 channel: Ex 540 nm/9, Em 570 nm/20; Cy5 channel: Ex 640 nm/9, Em 670 nm/ 20. Fluorescence data was normalized using the following equation: (F i À F 0 )/(F e À F 0 ) Â 100%. F i indicates the uorescence signal of measured samples. F 0 indicates the uorescence signal of probe-only negative controls. F e indicates the uorescence signal of probes in the presence of Exo I & III.

Escherichia coli strains and cell culture
Plasmids for mutants (D322N, A288V, S326C, S326E, S280E, S231E, and S232E) were provided by Prof. Dmitry O. Zharkov (Novosibirsk State University). These plasmids were transformed into E. coli BL21(DE3) competent cells using manufacturer protocols (New England BioLabs, USA). 29 Single colonies were cultured overnight at 37 C in LB broth (100 mg per mL ampicillin) and sequenced (Microsynth, Switzerland). The WT hOGG1 plasmid was constructed from a D322N mutant plasmid using a Q5 site-directed mutagenesis kit (New England BioLabs). PCR primers were designed using NEBaseChanger (New England BioLabs, USA): WTF, 5 0 -ATTGGCGCAGGTCGGCACTGAAC; WTR, 5 0 -TCCTCATATGAGGACTCTCGTAGCTGCTGCAG. Aer the PCR and ligation reaction, the plasmid was transformed into E. coli BL21(DE3) competent cells 29 and the cells were grown on a selection plate containing 100 mg per mL ampicillin. Several colonies were isolated and cultured. Plasmids were isolated from these colonies and sequenced (Microsynth, Switzerland). A WT hOGG1 plasmid conrmed by sequencing was used for further experiments. For protein expression, E. coli strains containing WT hOGG1 and mutants were cultured in LB broth (100 mg per mL ampicillin) overnight at 37 C. The starting culture was diluted with LB broth (100 mg per mL ampicillin) and shaken at 37 C until the cell density reached OD 600 0.4-0.6. Then isopropyl-b-D-thiogalactopyranoside (IPTG, nal concentration of 0.2 mM) was added, and incubation was continued at 30 C for 18 h. The cells were washed twice with 1Â PBS buffer and twice with 10% glycerol, and then centrifuged (7000g for 20 min at 4 C). The cell pellets were stored at À80 C for future use.

Cell lysate assays
All cell lysates were prepared from single clonal E. coli isolates treated with lysozyme (1 mg mL À1 ) and 1Â protease inhibitor (Roche, Switzerland) on ice for 30 min. Furthermore, the cell lysate assays were conducted in a similar fashion as with the puried proteins in the enzymatic assays using 10 6 cells. Lysates were added to a mixture of probes and buffer on ice, and the plate was allowed to equilibrate to 37 C for 2 min. Fluorescence data were acquired in the same manner as described above under enzymatic assays with molecular beacons (data shown in Fig. 5 and S8-S10 †).

Construction of site-directed D322 and C253 hOGG1 mutants
The D322 and C253 mutant library was created using a Q5 sitedirected mutagenesis kit (New England BioLabs, USA). PCR primers were designed using NEBaseChanger (New England BioLabs, USA): D322F, 5 0 -ATTGGCGCAGNNNGGCACTGAA-CAGC; D322R, 5 0 -CCCGCCATGCTCAGGAGC; C253F, 5 0 -TCAGGCAGATNNNGTCAGCCACCTTG; C253R, 5 0 -TGGCCCTA-GACAAGCCCC. Aer the PCR and ligation reaction, the plasmid library was transformed into E. coli BL21(DE3) competent cells and the cells were grown on a selection plate containing 100 mg per mL ampicillin. The resulting 96 colonies (48 from D322 and 48 from C253) were isolated and cultured in a deep well 96-well plate (Thermo Fisher Scientic, USA) overnight at 37 C in LB broth (100 mg per mL ampicillin). The starting culture was diluted with LB broth (100 mg per mL ampicillin) and shaken at 37 C until the cell density reached OD 600 0.4-0.6. Then IPTG (nal concentration of 0.2 mM) was added, and incubation was continued at 30 C for 18 h. The cells were washed twice with 1Â PBS buffer and twice with 10% glycerol and then centrifuged at 7000g for 20 min at 4 C. These cells were directly used for uorescence-based hOGG1 activity assays and sequenced by Sanger sequencing (Microsynth, Switzerland).